Directory UMM :Data Elmu:jurnal:A:Applied Soil Ecology:Vol13.Issue2.Oct1999:

Evaluation of the diversity of rhizobia in Brazilian

agricultural soils cultivated with soybeans

Heitor L.C. Coutinho

a,*

, ValeÂria M. Oliveira

b

,

Andrea Lovato

b

, Aline H.N. Maia

c

, Gilson P. Man®o

b

a_{Embrapa Solos, Rua Jardim BotaÃnico 1024, Rio de Janeiro-RJ, CEP 22640-000, Brazil} b_{Fundac,aÄo Andre Tosello, Rua Latino Coelho 1301, Campinas-SP, CEP 13087-010, Brazil}

c_{Embrapa Meio Ambiente, Rod. SP-340 Km. 127, JaguariuÂna-SP, CEP 13820-000, Brazil}

Received 20 May 1998; received in revised form 1 November 1998; accepted 10 March 1999

Abstract

The diversity of rhizobia in agricultural soils planted with soybean (Glycine maxL.) and managed under conventional or no-tillage practices was evaluated by using a combination of trap-host capture and DNA ®ngerprinting approaches. Fifty-eight rhizobia isolates were captured using pigeonpea (Cajanus cajanL.) as a trap-host and characterised by using the RAPD DNA ®ngerprinting technique, yielding 25 different RAPD pro®les. The application of the Shannon±Weaver diversity index demonstrated that the diversity of rhizobia was signi®cantly reduced in soil samples from plots cultivated with soybean compared with original uncultivated pasture plots.#1999 Elsevier Science B.V. All rights reserved.

Keywords: Rhizobium;Bradyrhizobium; Biodiversity; Agricultural soils; Tillage; DNA ®ngerprinting

1. Introduction

Rhizobia are nitrogen-®xing bacteria which usually develop symbiotic associations with leguminous plants. Recent taxonomic revision of this group of

organisms has de®ned ®ve different genera:

Azorhi-zobium,Bradyrhizobium,Mesorhizobium,Rhizobium andSinorhizobium(Young, 1996), taking into account phenotypic and phylogenetic characteristics of the organisms.

Studies of the diversity of rhizobia have been hampered by methodological dif®culties regarding sampling, identi®cation, and available methods for

environmental monitoring (Coutinho, 1996), related to an overall high diversity of this group of micro-organisms in nature (Laguerre et al., 1993; Moreira et al., 1993). Oyaizu et al. (1992) performed an extensive survey of 117 rhizobial strains isolated from 91 legume species, suggesting the existence of at least 16 species among the strains studied. Data from Moreira et al. (1993) demonstrated that the majority of rhizobia from soils of the Brazilian Atlantic and Amazonian rainforests were of the slow-growing, alkali-producing type, leading the authors to classify

them as Bradyrhizobium sp. Other researchers have

shown a high degree of diversity among rhizobia isolated from nodules of both woody and non-woody legumes (Haukka and LindstroÈm, 1994; Novikova et al., 1994; van Rossum et al., 1995). However, there are not many studies on the effects of agricultural

*Corresponding author. Tel.: 274-4999; fax: +55-021-274-5291

E-mail address:heitor@cnps.embrapa.br (H.L.C. Coutinho)

practices on the diversity of rhizobia in natural envir-onments in the tropics.

In this study, we assess the impact of different soil-management practices (conventional and no-tillage) on the diversity of rhizobia in plots planted with

soybean (Glycine maxL.), using the Shannon±Weaver

diversity index to analyse RAPD data derived from

strains captured with the legume pigeonpea (Cajanus

cajanL.) as trap-host.

2. Material and methods

2.1. Field experiment and soil samples

The ®eld experiment was set up in Fazenda Barce-lona (GuaõÂra, state of SaÄo Paulo, Brazil), in April 1995. Soils were of the acidic oxisol type, typical of the Brazilian Cerrado, and the area had been under pasture (Brachiaria decumbens) for several years prior to setting up the experiment. Three replicate plots,

918 m2each, were prepared by using conventional

tillage (CT), characterised by one mouldboard plough-ing of the soil followed by two disc-harrowplough-ing pro-cedures. The same number of replicate plots were prepared with no-tillage (NT), that is, the seeds and fertilisers were applied directly to the furrows opened over the residues from the previous crop. The plots

were sown with beans (Phaseolus vulgarisL.) (April±

July 1995) and left fallow until soybean (Glycine max

L.) was sown (December 1995), and managed accord-ing to the common agricultural practices in the region. Four composite soil samples, consisting of pooled samples taken from 10 randomly selected points at a depth of 20 cm, were collected prior to the prepara-tion of the pasture plots for planting (PO; April 1995). Thirty days after sowing soybean in the CT and NT plots (January 1996), composite soil samples were taken from each plot, with a total of three composite samples per treatment. The soil samples were refri-gerated at 48C during transport to the laboratory prior to the trap-host experiments.

2.2. Isolation procedures and phenotypic characterisation of strains

Rhizobia isolates were obtained from the soil sam-ples using pigeonpea plants (C. cajanL.) as trap-hosts.

This legume species was chosen due to its widespread utilisation as green manure in Brazilian agriculture. Capture experiments were performed in a greenhouse using Leonard jars and arti®cial sterile substrate (ver-miculite) in the upper compartment and plant nutrient solution in the lower compartment. Leonard jars setup and plant nutrient solution were performed as described previously (Somasegaran and Hoben, 1985). Three pre-germinated pigeonpea plants were

each inoculated with 2 ml of a 10ÿ2 soil dilution.

Thirty days after planting, three nodules per plant were harvested, surface-sterilised and macerated onto the surface of the growth media, as described pre-viously (Somasegaran and Hoben, 1985). Isolation

and growth media consisted of YMA (0.5 g KH2PO4;

0.2 g MgSO47H2O; 0.1 g NaCl; 0.5 g yeast extract;

10 g mannitol; 0.5% bromthymol blue solution; 15 g agar; per litre). The bacterial culture obtained around the nodule debris, after 3±7 days of incubation at 28₈C was streaked again over YMA plates for colony iso-lation. The isolates obtained were checked for purity after incubation at 288C, and their growth rates deter-mined. Strains were classi®ed as slow- (5±7 days) or fast- (3±4 days)growers according to their ability to

form single colonies of1±2 mm over the surface of

the culture media. Only one pure isolate was selected per nodule. Authentication of strains was performed using nodulation tests in pigeonpea plants, as described by Somasegaran and Hoben (1985). Isolates were preserved by storage of cell suspensions in 20% glycerol at ÿ20₈C (strains representative of distinct RAPD pro®les were also preserved by lyophilisation) and deposited at the Tropical Culture Collection (CCT, Fundac,aÄo Andre Tosello, Campinas±SP, Brazil).

2.3. Characterisation of strains by RAPD

Strains were characterised by using the RAPD technique (Coutinho et al., 1993). Rhizobial cells were

lysed in a solution containing 50 ngml proteinase K,

according to Young and Blakesley (1991), and their DNA concentration was estimated by agarose gel electrophoresis and ethidium bromide staining. The

25ml-volume reactions were prepared in duplicate,

each containing either 5 or 10 ng of rhizobial DNA,

2 U ofTaqDNA polymerase (CENBIOT, RS, Brazil),

2 mM MgCl2; 0.2 mM dNTPs and 1.0mM of a single

10-mer oligonucleotide primer (UBC#4, 50

GGG CTG G-30_{, University of British Columbia,}

Vancouver, BC). Cycling conditions were: 1 cycle

at 94₈C (2 min), followed by 30 cycles at 94₈C

(30 s), 36₈C (30 s) and 72₈C (1 min), with a ®nal

extension cycle at 72₈C (3 min). Agarose gel

electro-phoresis was performed according to Sambrook et al. (1989), using 1.2% gels. RAPD pro®les were regis-tered on Polaroid ®lms after ethidium bromide stain-ing of gels.

2.4. Hybridisation with Bradyrhizobium sp./B. elkani probe

ABradyrhizobiumsp./B. elkani-speci®c probe (50

-CCg TCT CTg gAg TCC gCg ACC-30_{), designed by}

Oliveira et al. (1997), was used to check the identity of some of the soil isolates. This probe was shown to

hybridise with DNA from Bradyrhizobium sp. and

B. elkani, but not to Bradyrhizobium japonicum. Genomic DNA from pure cultures of rhizobia was puri®ed according to Pitcher et al. (1989). 16S rDNA was then ampli®ed by using universal primers p27f

and p1401r (Lane, 1991). PCR reactions (50ml)

con-tained 50 ng of genomic DNA, 2 U of Taq DNA

polymerase (CENBIOT, RS, Brazil), 2 mM MgCl2;

0.2 mM dNTPs and 0.4mM of each primer.

Ampli®-cation conditions consisted of one cycle at 958C

(2 min), followed by 30 cycles at 94₈C (1 min),

60₈C (1 min) and 72₈C (3 min), with a ®nal extension

cycle at 72₈C (5 min), in a Perkin±Elmer 9600 thermal

cycler. Ampli®ed 16S rDNA were electrophoresed on 1.2% agarose gels, denatured and vacuum-transferred (VacuGene XL-Pharmacia) to nylon membranes (Boehringer±Mannheim), according to the

manufac-turer's instructions. TheBradyrhizobiumsp./B. elkani

probe was 30

labelled with digoxigenin (DIG-11-dUTP) using the terminal transferase enzyme (GIBCO-BRL), according to the manufacturer's

instructions. Labelling reactions (20ml) contained

50 U of terminal transferase, 12mM oligonucleotide,

0.5 mM dATP and 0.05 mM DIG-11-dUTP (Boehrin-ger±Mannheim) in 1X reaction buffer. The labelling reaction was carried out overnight at 37₈C.

Hybridisa-tions were carried at 50₈C for 18 h, followed by two

washes with 6X SSC for 15 min at room temperature and one stringent wash at 60₈C for 3 min. Probe signal was detected by chemiluminescence (DIG Lumines-cent Detection Kit, Boehringer±Mannheim), using

CSPD, and recorded on X-ray ®lm by exposure for 1 min.

2.5. Data analysis and evaluation of diversity

The rhizobia isolates were grouped according to the degree of similarity of their RAPD banding patterns. The Shannon±Weaver index (Shannon and Weaver, 1949) was applied to determine the diversity index (H) for each treatment, and was calculated by the follow-ing equation:

Hˆ ÿX

k

iˆ1 pilnpi

where k is the number of operational taxonomic

units (OTU; de®ned by similar RAPD pro®les) and

pi, the relative abundance of isolates of each OTU

(iˆ1,2,. . .,k). An adaptation of the Pielou's pooled

quadrate method (Pielou, 1969), consisting of 300 random ordinations of the data and calculation of the diversity indices in a cumulative manner, was applied. Con®dence limits were determined by

calculat-ing the jackknife estimate of the variance ofH

(Magur-ran, 1988).

3. Results

3.1. Isolation and phenotypic characterisation of strains

A total of 58 strains were isolated from pigeonpea nodules taken from plants cultivated in substrates inoculated with soil samples from pasture, conven-tional tillage and no-tillage plots. All putative rhizo-bial strains isolated were able to form nodules in

pigeonpea (C. cajanL.), thus characterising them as

rhizobia. Strains which formed nodules were further classi®ed as slow- and fast-growers according to their growth rates (Table 1).

3.2. DNA fingerprinting of rhizobial isolates

Table 1

Classification of rhizobial isolates according to their growth rates, RAPD banding patterns and hybridisation signal with aBradyrhizobiumsp./ B.elkani-specific probe

Soil sample Isolate Growth rate

OTUa,b _{Hybridisation signal with}

Bradyrhizobiumsp./B.elkaniprobec

Pasture (P0)

CCT 6186 slow SMC ‡

CCT 6187 slow I ‡

CCT 6188 slow I ND

CCT 6189 slow SMC ‡

CCT 6190 slow I ND

CCT 6191 slow SMC ÿ

CCT 6192 slow SMC ND

CCT 6193 slow SMC ÿ

CCT 6194 slow IV ‡

CCT 6195 slow IV ND

CCT 6196 slow IV ND

CCT 6197 slow SMC ÿ

CCT 6198 slow IV ND

CCT 6199 slow SMC ‡

CCT 6200 slow SMC ‡

CCT 6201 slow SMC ‡

CCT 6202 slow SMC ND

CCT 6203 slow SMC ‡

CCT 6204 slow III ‡

CCT 6205 slow V ‡

CCT 6206 slow V ND

CCT 6207 slow SMC ‡

CCT 6208 slow I ND

CCT 6209 slow I ND

CCT 6210 slow I ND

CCT 6211 slow I ND

CCT 6212 slow III ‡

CCT 6213 slow SMC ‡

CCT 6214 slow SMC ‡

CCT 6215 slow SMC ‡

CCT 6216 slow SMC ‡

Conventional tillage (CT)

CCT 6217 fast II ÿ

CCT 6218 fast II ND

CCT 6219 slow I ND

CCT 6220 slow VII ‡

CCT 6221 slow I ND

CCT 6222 fast VIII ÿ

CCT 6223 slow I ‡

CCT 6224 fast VIII ND

CCT 6225 slow VII ND

CCT 6226 slow I ND

CCT 6227 fast II ND

CCT 6228 slow VII ND

CCT 6229 fast VI ÿ

CCT 6230 slow I ND

CCT 6231 slow IX ‡

diverse group of isolates, which included 16 organ-isms with unique RAPD pro®les. OTU I isolates were found in all soil samples. OTUs III, IV, and V were found solely in the pasture soil. Likewise, OTUs II, VI,

VII, VIII, and IX were detected only in soil samples from the cultivated plots. Fast-growing rhizobia from group VI prevailed among isolates from soil samples of the no tillage (NT) treatment.

Table 1 (Continued)

Soil sample Isolate Growth rate

OTUa,b _{Hybridisation signal with}

Bradyrhizobiumsp./B.elkaniprobec

No tillage (NT)

CCT 6233 slow I ND

CCT 6234 fast VI ND

CCT 6235 slow SMC ‡

CCT 6236 fast VI ND

CCT 6237 fast VI ND

CCT 6238 fast VI ND

CCT 6239 fast SMC ÿ

CCT 6240 slow I ND

CCT 6241 fast VI ND

CCT 6242 fast SMC ÿ

CCT 6243 fast VI ND

a_{(OTU) Operational taxonomic units: named I to IX (see text for details).} b_{(SMC) Single-membered cluster: isolate with unique banding pattern.} c_{(ND) Not determined.}

3.3. Analysis of diversity

The Shannon±Weaver diversity index (H) was

cal-culated, and the corresponding curves constructed, showing its variation as a function of the number of isolates considered in the analysis (Fig. 2). For treat-ments CT (conventional tillage) and NT (no-tillage), the diversity-index curves stabilised at nodule num-bers >10. Treatment P0 comprised data from 31 strains and its curve followed the overall pattern of the others. However, the diversity-index values increased

gradu-ally with isolate numbers >16, suggesting that theH

values obtained are underestimates and more isolates are needed. Nevertheless, the rhizobial diversity in the P0 soil was greater than that in the cultivated plots. The Shannon±Weaver index jackknife estimates demonstrate that the difference between the index for P0 and that for the cultivated plots is statistically signi®cant (Table 2 and Fig. 3). No signi®cant

differ-ence was found between the diversity indices of the CT and NT treatments.

4. Discussion

The use of pigeonpea (C. cajan L.) to capture

rhizobia proved to be an ef®cient strategy. Its low host-speci®city was demonstrated by the diversity of isolates recovered. Rhizobia of both slow- and fast-growing types were isolated, indicating that species of different genera were recovered (Table 1). Some of the slow-growing isolates could be identi®ed as

belonging to the genus Bradyrhizobium, and were

putatively assigned toBradyrhizobium sp. orB.elkani

by their hybridisation signal with theBradyrhizobium

sp./B. elkaniprobe.

No fast-growing rhizobia were recovered from the P0 soil samples (prior to planting of soybean). On the

Fig. 2. Shannon±Weaver cumulative diversity curve for the different agricultural treatments: means ofHindices as a function of sample size. P0, pasture; CT, conventional fillage; and NT, no tillage.

Table 2

Jackknife estimate of the Shannon±Weaver diversity indices (Hj) for the different agricultural treatments

Treatment n Shannon±Weaver diversity indices (Hj)

mean SD confidence

limit (lower)

confidence limit (upper)

Pasture (P0) 31 3.30 0.22 2.84 3.76

Conventional tillage (CV) 16 1.90 0.16 1.55 2.24

other hand, the majority of isolates from the NT samples (no tillage) were fast-growers. Clearly, the different soil-management practices produced altera-tions in the pro®le of the rhizobial populaaltera-tions in the soil. However, the trends observed were contrary to the expectation that the soybean crop should favour an increase of bradyrhizobia.

The number of operational taxonomic units (OTUs) and the number of individuals belonging to each OTU (`richness' and `abundance', respectively) were used to calculate the Shannon±Weaver diversity index (Shannon and Weaver, 1949) of each treatment. The P0 soil samples presented a signi®cantly higher rhi-zobial diversity when compared to that of NT and CT (Table 2; Figs. 2 and 3). In fact, 16 out of 31 isolates recovered from P0 produced unique RAPD ®nger-prints (Table 1). This may explain why the diversity index curve for this treatment did not stabilise, unlike treatments NT and CT (Fig. 2). As the rhizobial diversity in a soil increases, a larger sample size, i.e., more isolates, should be analysed in order to achieve a more precise estimate of their diversity.

The variance of the diversity-index estimates was calculated using the `jackknife' procedure, which

provided a statistical signi®cance test to the compara-tive analysis undertaken. The jackknives of the

diver-sity index (Hj) of treatments CT and NT were

statistically identical (Table 2) and different from those presented in Fig. 2. This could be explained by the fact that the jackknife has a bias of the order 1/n (Quenouille, 1956).

The changes in rhizobial diversity observed in soils converted to soybean cultivation are probably a result of the profound modi®cations imposed on the soil environment by the application of agrochemicals (fertilisers, herbicides, and pesticides), by physical disturbance (ploughing and tillage), and by the soy-bean crop itself. One does not know whether altera-tions in rhizobial diversity will affect the nodulation potential of a soil. In order to address these matters, further investigation should focus on the symbiotic performance of the rhizobial strains isolated in this study.

Although diversity evaluations based on isolated rhizobia may be biased, since only those strains able to nodulate the legume species used for capture are analysed, this approach proved to be useful for com-parative analyses, such as the one presented in this

work. This study could be complemented by an ana-lysis of the sequence diversity of rhizobial gene frag-ments ampli®ed from total bacterial DNA extracted from the soil samples (Stackebrandt et al., 1993). PCR primers or DNA probes, speci®c to different rhizobial taxa, are expected to be developed in the near future, which will enable the evaluation of the diversity of a broader range of rhizobia in soil, most of them likely to be unknown. However, conventional isolation and characterisation methods will still be important, since they enable subsequent screening for novel legume seed inoculants.

The data obtained and the statistical framework adopted in the current study is, to our knowledge, pioneering in the study of rhizobia diversity. This methodology will be applicable to a variety of com-parative studies based on molecular data of micro-organisms. It is currently being applied to evaluate how much each individual host plant in¯uences the selection of nodulating strains. This has been shown to

greatly in¯uence population-diversity studies of

Rhi-zobium leguminosarum biovarviciae in British soils (Handley et al., 1998).

5. Conclusions

A methodology for comparative analysis of rhizo-bial diversity in soils has been proposed. It was used to

evaluate how the diversity of pigeonpea (C. cajanL.)

rhizobia is affected by different types of land use and tillage practices in a tropical soil in the Brazilian cerrado. The results indicate that conversion of the

land use from pasture (Brachiaria decumbens) to

soybean (Glycine maxL.) led to a signi®cant decrease

in the diversity of pigeonpea rhizobia. No statistical difference was found between the Shannon±Weaver diversity indices of rhizobia in soils prepared by no-till and conventional tillage practices. However, DNA ®ngerprinting showed that the strains recovered from soils under no-tillage were different from those under conventional tillage. The determination of growth

rates and hybridisation with a [Bradyrhizobium sp./

B.elkanii]-speci®c DNA probe demonstrated that the majority of pigeonpea nodule bacteria recovered from the soils prepared by conventional tillage or under

pasture belong to theBradyrhizobiumgenus. The use

of the jackknife procedure enabled statistical analysis

of the Shannon±Weaver diversity indices, in spite of the low number of isolates studied. However, higher numbers of isolates and advanced molecular systema-tics procedures, such as soil DNA extraction and sequence analysis, are required to allow the observa-tion of changes in rhizobial populaobserva-tion structures in soils under different types of land use.

Acknowledgements

H.L.C. Coutinho and V.M. Oliveira were supported by a CNPq Research Grant and FAPESP Doctoral Grant, respectively. Presentation of results in the XVI World Congress of Soil Science was sponsored by CNPq and FAPERJ. The authors wish to thank Dr. Pedro Valarini for the support to set up the ®eld experiment and Ms. Carla M. Soares and Mr. Jefferson Mineiro for valuable technical assistance.

References

Coutinho, H.L.C., 1996. Diversidade microbiana e agricultura sustentaÂvel. In: Workshop: BiodiversidadeÐPerspectivas e Oportunidades TecnolOÂ gicas. On-line publication, http:// www.bdt.org.br/bdt/paper/padctbio/cap9/.

Coutinho, H.L.C., Handley, B.A., Kay, H.E., Stevenson, L., Beringer, J.E., 1993. The effect of colony age on PCR fingerprinting. Lett. App. Microbiol. 17, 282±284.

Handley, B.A., Hedges, A.J., Beringer, J.E., 1998. The importance of host plants for detecting the population diversity of Rhizobium leguminosarum biovar viciae in soil. Soil Biol. Biochem. 30(2), 241±249.

Haukka, K., LindstroÈm, K., 1994. Pulse-field gel electrophoresis for genotypic comparison ofRhizobiumbacteria that nodulate leguminous trees. FEMS Microbiol. Lett. 119, 215±220. Lane, D.J., 1991, 16S/23S rRNA sequencing. In: Goodfellow,

M., Stackebrandt, E. (Eds.), Nucleic Acid Techniques in Bacterial Systematics. John Wiley & Sons, Chichester. pp. 115±147.

Laguerre, G., Geniaux, E., Mazurier, S., Rodriguez-Casartelli, R., Amarger, N., 1993. Conformity and diversity among field isolates ofRhizobium leguminosarumbyviciae, byTrifoliiand byphaseolirevealed by DNA hybridisation using chromosome and plasmid probes. Can. J. Microbiol. 39(4), 412±419. Magurran, A.E., 1988, Ecological Diversity and its Measurement.

Croom Helm, London.

Novikova, N.I., Pavlova, E.A., Vorobjev, N.I., Limenshchenko, E.V., 1994. Numerical taxonomy ofRhizobium strains from legumes of the temperature zone. Int. J. Syst. Bacteriol. 44, 734±741.

Oliveira, V.M., Rosato, Y.B., Coutinho, H.L.C., Manfio, G.P., 1997, Design of a 16S rRNA-directed oligonucleotide probe for Bradyrhizobiumtropical strains. In: Elmerich, C., Kondorosi, A., Newton, W.E. (Eds.), Biological Nitrogen Fixation for the 21st Century. Proceedings of the 11th International Congress on Nitrogen Fixation, Paris, France. Kluwer Academic Publishers, Dordrecht. pp. 581.

Oyaizu, H., Naruhashi, N., Gamou, T., 1992. Molecular methods of analysing bacterial diversity: the case of Rhizobia. Biodiv. Conserv. 1, 237±249.

Pielou, E.C., 1969. An Introduction to Mathematical Ecology. Wiley, New York.

Pitcher, D.G., Saunders, N.A., Owen, R.J., 1989. Rapid extraction of bacterial genomic DNA with guanidium thiocyanate. Lett. Appl. Microbiol. 8, 151±156.

Quenouille, M.H., 1956. Notes on bias in estimation. Biometrika 43, 353±360.

Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory Manual. Cold spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.

Shannon, C.E., Weaver, W., 1949. The Mathematical Theory of Communication. University of Illinois Press, Urbana. Somasegaran, P., Hoben, H.J., 1985. Methods in legume±

Rhizobium technology. University of Hawaii NifTAL Project and MIRCEN, USA.

Stackebrandt, E., Liesack, W., Goebel, B.M., 1993. Bacterial diversity in a soil sample from a subtropical Australian environment as determined by 16S rDNA analysis. FASEB J. 7, 232±236.

van Rossum, D., Schuurmans, F.P., Gillis, M., Muotcha, A., van Verseveld, H.W., Stouthamer, A.H., Boogerd, F.C., 1995. Genetic and phenetic analyses of Bradyrhizobium strains nodulating peanut (Arachis hypogaeaL.) roots. Appl. Environ. Microbiol. 61, 1599±1609.

Young, J.P.W., 1996. Phylogeny and taxonomy of rhizobia. Plant Soil 186, 45±52.